What it shows. Each dot is a Mexican restaurant, taqueria, or taco spot in Santa Clara or San Mateo County. For each one it compares the standard Google star rating with an authenticity-weighted rating that counts reviews from likely Hispanic/Latino reviewers twice.
A review is flagged if it was written in Spanish, or if the reviewer’s surname has a high Hispanic share in the U.S. Census Bureau’s 2010 surname file (≥75% of people with that surname self-identify as Hispanic) — the same public data basis used in fair-lending analysis. It is a population-level probability, not a claim about any individual’s identity or heritage.
Up to the 50 most recent public Google reviews per restaurant (~29,000 reviews across 682 places). Ratings shown are computed from that sample, not Google’s lifetime average.
The Clientele-weighted layer multiplies the reviewer-weighted rating by each restaurant’s real Hispanic patronage: weighted★ × (1 + visitor Hispanic share). Visitor Hispanic share comes from Advan/SafeGraph mobility data (2025) — the Census-block-group home locations of actual visitors, weighted by Census (ACS 2024 5-yr) Hispanic share. Covers ~547 restaurants with sufficient visitor data; others show grey. The × review volume variant additionally multiplies by a log-scaled factor of each restaurant’s total Google reviews (1–2×), rewarding established spots without letting high-volume chains dominate.
This is a beta and an exploratory research tool. The flag is a noisy proxy: only ~71% of reviewer names yield a usable surname, and short reviews weaken the language signal. Treat it as an aggregate signal, never as a label for a person.
After correcting for testing hundreds of restaurants at once, the flagged-vs-other rating gap is individually significant for only 25 of 617 restaurants at q<0.005 (the green rings). For every other dot, please don’t over-read a single restaurant’s gap — the meaningful signal is the overall tendency, not any one place.
Not affiliated with or endorsed by Google. Built from publicly visible reviews for research and education. Surname data: U.S. Census Bureau, Frequently Occurring Surnames from the 2010 Census.